Research On Algorithm Of Collision Detection Based On Bounding Boxes

Collision detection is not only an important research direction in real-time computer graphics but also a key technology in a variety of areas such as computer-aided design, computer animation, and virtual reality. Efficient and accurate collision detection is very important to improve reality and enhance immersion in these mentioned applications. Based on hybrid hierarchical bounding volumes collision detection algorithm and computer graphics hardware technology, two aspects have been researched to improve the efficiency of collision detection algorithm.Using different types of bounding boxes to detect intersections, hybrid hierarchical bounding volumes collision detection algorithm has provided an effective way to improve the efficiency of the collision detection algorithm that using only one type. The diversity of the bounding boxes causes the diversity of hybrid hierarchical bounding volumes. So methods of calculating the intersections between several different common types of bounding boxes have been studied, then three factors influencing the shape of hybrid hierarchical bounding volumes have been analyzed, and finally a method of building hybrid hierarchical bounding volumes has been presented.Although collision detection using hybrid hierarchical bounding volumes can improve the calculations effectively, the key calculations are done on CPU, the computing power of GPU can not be brought into full play, causing a waste of computing resources; the streaming pattern collision detection algorithm based on computer graphics hardware technology has a good parallel processing capabilities, but requires too much unnecessary detection, which affects the efficiency. After analyzing the characteristics of these two collision detection algorithm, a hybrid collision detection algorithm has been presented. Firstly, the hybrid hierarchical bounding volume collision detection is used to find the precise location where intersection may occur and then using streaming pattern collision detection to check out whether collision actually happens. In this way, the advantages of the original two algorithms are balanced. Experiments show that the efficiency of the improved algorithm was increased than the original algorithm. Besides, the improved algorithm also has a good adaptability to complex scenes.